We benchmarked 18 LLMs on OCR (7K+ calls)
cheaper models win
Details
- External ID
- 47860393
- Source
- HN
- Company
- —
- Product
- We benchmarked 18 LLMs on OCR (7K+ calls)
- Website domain
- arbitrhq.ai
- Launched
- April 22, 2026
- Cohort
- —
- Upvotes
- 5
- Upvotes percentile
- 0.11182519280205655
- Tags
- —
- Fetched at
- Sept. 7, 2026, 9:26 p.m.
- Updated at
- Sept. 7, 2026, 9:26 p.m.
Enrichment
- Theme
- ML inference and model optimization
- Vertical
- Horizontal
- Function
- Analytics & BI
- Audience
- Developer
- AI stance
- Not AI
- Project type
- Commercial product
- Normalized one-liner
- llm ocr benchmark
- Manually corrected
- False
Could you build this?
Yes This is an evaluation and benchmarking harness that passes sample documents through various commercial LLM APIs and records the accuracy and latency on a static leaderboard.
Discussion
1 comment analyzed.
Competitors mentioned: OpenAI models, Anthropic models, Google models, Mistral models, Legacy OCR pipelines
Concerns raised: Teams overpaying for newest/biggest models, Reliability degradation over repeated runs, Edge cases where flagship models may still be necessary
Feature requests: Edge case documentation for when flagship models justify cost, Comparison with legacy OCR solutions on same benchmark
Competitors
Other products that read as similar to this one — 204 launches clear the similarity bar, closest 8 shown.
Attention rank: #176 of 205 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).
Launched 174 days after the earliest competitor.
- We built an OCR server that can process 270 dense images/s on a 5090 · hn · 2026-04-23 · 8 upvotes · similarity 0.53
- 17MB model beats human experts at pronunciation scoring · hn · 2026-02-20 · 13 upvotes · similarity 0.52
- LLM Debate Benchmark · hn · 2026-03-23 · 9 upvotes · similarity 0.49
- Rapid-MLX · hn · 2026-04-18 · 9 upvotes · similarity 0.48
- oido · github · 2026-09-30 · 24 upvotes · similarity 0.47
- Minimal LLM Post-Training Experiments on an 8GB GPU (SFT, DPO, GRPO) · hn · 2026-08-01 · 21 upvotes · similarity 0.46
- CPU-only fast OCR for screenshots, images, PDFs, webpages · hn · 2026-05-31 · 9 upvotes · similarity 0.46
- CLaaS · hn · 2026-02-26 · 6 upvotes · similarity 0.46
Other launches for this product
- No other launches for this product.
Same idea, different domain
Nobody's really built a analytics & bi tool for Legal yet.